KO | EN

GitLite

Minimalist Open Source Repository Search Engine

gitlite — search

#dsh #dsh-plugin #ai-agents #developer-tools #typescript #codex #cli #claude-code #mcp #rust #deepseek-harness #ai-agent

raman_tl

★ 29 Open GitHub ↗

Baseline correction, smoothing, processing and plotting of Raman spectra

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# raman-tl.py

> [!TIP]
> **Trusty `raman-tl.py` is now available as a web app (thanks to AI)! Enjoy a brand-new interface.**  
> 👉 Try it here: https://advancedramantl-nfvgsz7dquxrmtk6xvc9hq.streamlit.app  
> 👉 Dive into the [code](https://github.com/radi0sus/advanced_raman_tl).

A Python 3 script for baseline correction, smoothing, processing and plotting of Raman spectra. Data must be in the format `wavenumber [space] intensity`. The baseline correction uses the *asymmetrically reweighted penalized least squares smoothing algorithm* (arPLS). The Whittaker filter is (by default) applied for smoothing. Optionally, the [Savitzky–Golay](https://en.wikipedia.org/wiki/Savitzky–Golay_filter) filter can be used. Data of the processed spectra can be saved as "csv"-like data files in the format `wavenumber [delimiter] intensity`. An overlay spectrum (normalized and not normalized) and a normalized stacked spectrum of all processed spectra can be plotted. Plots can be saved as PNG bitmap files and as PDF.

If you use the arPLS algorithm to process your spectra, please cite:

> "Baseline correction using asymmetrically reweighted penalized least squares smoothing"
> 
> Sung-June Baek, Aaron Park, Young-Jin Ahna, Jaebum Choo, *Analyst* **2015**, *140*, 250-257
> 
> DOI:	https://doi.org/10.1039/C4AN01061B

The Whittaker algorithm (sometimes also referred to as Whittaker-Eilers smoother) is adapted from:

> "A perfect smoother"
> 
> Paul H. C. Eilers, *Anal. Chem.* **2003**, *75*, 3631-3636
> 
> DOI https://doi.org/10.1021/ac034173t

which is based on:

>"On a new method of gradutation"
>
>E. T. Whittaker, *Proceedings of the Edinburgh Mathematical Society* **1922**, *41*, 63-75
>
>DOI: https://doi.org/10.1017/S0013091500077853

## External modules
 `numpy`,  `scipy`,  `matplotlib`
 
## Quick start
 Start the script with:
```console
python3 raman-tl.py filename
```
to open a single file.

Start the script with:
```console
python3 raman-tl.py *.txt
```
to process all files with the extension  `.txt` in the folder.

Under Windows, you have to open `PowerShell` first and start the script with:
```console
python raman-tl.py (Get-ChildItem *.txt -Name)
```
to process all files with the extension  `.txt` in the folder.

*If the plot window appears empty, please resize it. Check "Known issues" for further details.*

In all cases a file `summary.pdf` will be created which contains the following plots:

On the first page (from top to bottom):
- raw spectrum with baseline plot (red)
- baseline corrected spectrum
- smoothed / filtered spectrum with peak annotation

On the following page(s):
- smoothed / filtered spectrum with peak annotation 
- not normalized and normalized overlay spectra and normalized stacked spectra if the `-o` option was invoked

## Command-line options
- `filename` , required: filename(s), input file(s) in the format `wavenumber [space] intensity`
- `-l` `N`, optional: the lambda parameter for the arPLS algorithm (default is `N = 1000`)
- `-p` `N:M`, optional: invokes the Savitzky–Golay filter, `N:M` are the window length and polynomial order of the Savitzky–Golay filter 
- `-w` `N`,  optional: the lambda parameter for the Whittaker filter (default is `N = 1`)
- `-xmin` `N` , optional: start spectra at `N` wave numbers
- `-xmax` `N` , optional: end spectra at `N` wave numbers
- `-t` `N` , optional: threshold for peak detection, with `N` being the intensity (default is 5% from the maximum intensity)
- `-m` `N` , optional: multiply intensities with `N` (default is `N = 1`)
- `-a` `N` , optional: add or subtract `N` to / from wave numbers (default is `N = 0`)
- `-i` `N` , optional: add or subtract `N` to / from intensities (default is `N = 0`)
- `-o` , optional: show the normalized and not normalized overlay spectrum and the normalized stacked spectrum
- `-n` , optional: do not save `summary.pdf`
- `-s` `p,d` , optional: save P(NG) and / or D(ATA) files. The filenames are `filename.png` and / or `filename-mod.dat` for the single spectra. Data files are in the format `wavenumber [delimiter] intensity`. The delimiter can be set in the script. The default delimiter is [space]. `Summary.png`, `overlay.png`, `overlay-normalized.png`, `stack-normalized.png` bitmaps will be saved as well, overlay and stacked spectra only if the `-o` option has been invoked.

## Remarks
- The save values for the arPLS parameter `lambda` start from 1000. Smaller values will give sharper peaks, but broader peaks become part of the baseline. Check the red baseline curve in the summary page.
- There is no way to turn off smoothing directly, but with two Savitzky-Golay parameters close together, e.g. `-p3:2` or a Whittaker parameter `-w0.01` filtering is ineffective. 
- The window length for the Savitzky–Golay filter must be an odd number and the window length must be greater than the polynomial order.
- Polynomial based filters, such as the Savitzky–Golay filter, sometimes tend to overshoot in negative regions, especially with sharp signals in the Raman spectrum. Reduce the filtering (see above) is one way to solve this problem. 
- `xmin` and or `xmax` values outside the experimental wave number range will result in errors or strange outputs.
- `-a` changes the range for `xmin` and `xmax`
- `-i` and `-m` change the range for  `-t` 
- The  `.dat` file contains the data of the processed spectrum in the given range as it is shown in the plot for the single spectrum.
- The `-o` option invokes the overlay plots (normalized and not normalized) and the normalized stacked plot of all processed spectra. Normalized means, that the intensities are divided by the maximium intensity in the given intensity range. The maximum intensity becomes unity. The peak detection threshold for the normalized spectrum is 0.05 (can be changed in the script: `normalized_height`).
- The delimiter  in the `.dat` file can be changed in the script: `dat_delimiter = " "` or `dat_delimiter = " ; "` for example.
- The files `summary.pdf`, `summary.png`, `overlay.png`, `overlay-normalized.png`, `stack-normalized.png` will be overwritten every time the script is started (with respective options) in the same directory. Single spectra with the same filenames will be overwritten as well. Rename them if you want to keep them.

## Known issues
- Some of the peaks that are close together are not annotated. To change this, one can reduce the `peak_distance` in the script, which is by default `peak_distance = 8`.
- Peak annotations can be overprinted by other peak annotations in the overlay spectrum. There is no workaround for this. If annotations are in the same position, one can uncomment the instruction under `#no dupes` in the script, then only one annotation is displayed. 
- The legend obscures part of the spectrum. If this is a problem, one can change the position of the legend in the script or prevent the legend from being printed at the spectrum (try to change `head_space_y_o_s` in the script for the overlay and stacked spectra).
- Recent versions of Matplotlib and Python may encounter an issue where the plot window appears empty. As a temporary solution, resizing the window seems to solve the problem. This issue is currently unresolved: https://github.com/matplotlib/matplotlib/issues/25768. However, with Matplotlib version 3.9.0, everything functions as expected.

## Examples
![show](https://raw.githubusercontent.com/radi0sus/raman_tl/main/examples/show-use4.gif)

Remember, under Windows you have to open `PowerShell` first and start the script with:
```console
python raman-tl.py (Get-ChildItem *.txt -Name)
```
to open more than one file at once.

### Example 1
```console
python3 raman-tl.py s*.txt
```
Process all files starting with `s` and the extension `.txt`.

Summary:

Single spectra:

### Example 2
```console
python3 raman-tl.py sample-A.txt -xmin 600 -xmax 800 -spd 
```
Process spectrum `sample-A.txt` in the range from `xmin = 600` to `xmax = 800` cm-1 and save the PNG and DATA files (`-spd`).

Summary:

Single spectrum:

### Example 3
```console
python3 raman-tl.py sample-A.txt -l10000 -p7:4 -xmin 600 -xmax 800 -t50 -spd 
```
Process spectrum `sample-A.txt` with `lambda = 10000` (baseline parameter), `window length = 7` and `polynomial order = 4` (smoothing parameters) in the range from `xmin = 600` to `xmax = 800` cm-1, annotate peaks with intensities equal or greater than `t = 50` and save the PNG and DATA files (`-spd`).

Summary:

Single spectrum:

### Example 4
```console
python3 raman-tl.py sample-A.txt sample-B.txt -o -xmin 200 -xmax 1100 -sp
```
Process spectra `sample-A.txt` and `sample-B.txt` in the range from `xmin = 200` to `xmax = 1100` cm-1, plot the overlay and stacked spectra (`-o`) and save the PNG files (`-sp`).

Overlay spectrum (not normalized):

Overlay spectrum (normalized):

Stacked spectrum (normalized):

raman_tl

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README